]> git.proxmox.com Git - ceph.git/blobdiff - ceph/src/boost/boost/multiprecision/cpp_int/import_export.hpp
import new upstream nautilus stable release 14.2.8
[ceph.git] / ceph / src / boost / boost / multiprecision / cpp_int / import_export.hpp
index b975a79663f5f3147e45dcea4daf45d698d9bc8f..98b533090ccdd16c5d09988036132dc4e2b4faf3 100644 (file)
 ///////////////////////////////////////////////////////////////
 //  Copyright 2015 John Maddock. Distributed under the Boost
 //  Software License, Version 1.0. (See accompanying file
-//  LICENSE_1_0.txt or copy at http://www.boost.org/LICENSE_1_
+//  LICENSE_1_0.txt or copy at https://www.boost.org/LICENSE_1_0.txt
 
 #ifndef BOOST_MP_CPP_INT_IMPORT_EXPORT_HPP
 #define BOOST_MP_CPP_INT_IMPORT_EXPORT_HPP
 
-
 namespace boost {
-   namespace multiprecision {
-
-      namespace detail {
-
-         template <class Backend, class Unsigned>
-         void assign_bits(Backend& val, Unsigned bits, unsigned bit_location, unsigned chunk_bits, const mpl::false_& tag)
-         {
-            unsigned limb = bit_location / (sizeof(limb_type) * CHAR_BIT);
-            unsigned shift = bit_location % (sizeof(limb_type) * CHAR_BIT);
-
-            limb_type mask = chunk_bits >= sizeof(limb_type) * CHAR_BIT ? ~static_cast<limb_type>(0u) : (static_cast<limb_type>(1u) << chunk_bits) - 1;
-
-            limb_type value = static_cast<limb_type>(bits & mask) << shift;
-            if(value)
-            {
-               if(val.size() == limb)
-               {
-                  val.resize(limb + 1, limb + 1);
-                  if(val.size() > limb)
-                     val.limbs()[limb] = value;
-               }
-               else if(val.size() > limb)
-                  val.limbs()[limb] |= value;
-            }
-            if(chunk_bits > sizeof(limb_type) * CHAR_BIT - shift)
-            {
-               shift = sizeof(limb_type) * CHAR_BIT - shift;
-               chunk_bits -= shift;
-               bit_location += shift;
-               bits >>= shift;
-               if(bits)
-                  assign_bits(val, bits, bit_location, chunk_bits, tag);
-            }
-         }
-         template <class Backend, class Unsigned>
-         void assign_bits(Backend& val, Unsigned bits, unsigned bit_location, unsigned chunk_bits, const mpl::true_&)
-         {
-            typedef typename Backend::local_limb_type local_limb_type;
-            //
-            // Check for possible overflow, this may trigger an exception, or have no effect
-            // depending on whether this is a checked integer or not:
-            //
-            if((bit_location >= sizeof(local_limb_type) * CHAR_BIT) && bits)
-               val.resize(2, 2);
-            else
-            {
-               local_limb_type mask = chunk_bits >= sizeof(local_limb_type) * CHAR_BIT ? ~static_cast<local_limb_type>(0u) : (static_cast<local_limb_type>(1u) << chunk_bits) - 1;
-               local_limb_type value = (static_cast<local_limb_type>(bits) & mask) << bit_location;
-               *val.limbs() |= value;
-               //
-               // Check for overflow bits:
-               //
-               bit_location = sizeof(local_limb_type) * CHAR_BIT - bit_location;
-               if((bit_location < sizeof(bits)*CHAR_BIT) && (bits >>= bit_location))
-                  val.resize(2, 2); // May throw!
-            }
-         }
-
-         template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator>
-         inline void resize_to_bit_size(cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& newval, unsigned bits, const mpl::false_&)
-         {
-            unsigned limb_count = static_cast<unsigned>(bits / (sizeof(limb_type) * CHAR_BIT));
-            if(bits % (sizeof(limb_type) * CHAR_BIT))
-               ++limb_count;
-            static const unsigned max_limbs = MaxBits ? MaxBits / (CHAR_BIT * sizeof(limb_type)) + ((MaxBits % (CHAR_BIT * sizeof(limb_type))) ? 1 : 0) : (std::numeric_limits<unsigned>::max)();
-            if(limb_count > max_limbs)
-               limb_count = max_limbs;
-            newval.resize(limb_count, limb_count);
-            std::memset(newval.limbs(), 0, newval.size() * sizeof(limb_type));
-         }
-         template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator>
-         inline void resize_to_bit_size(cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& newval, unsigned, const mpl::true_&)
-         {
-            *newval.limbs() = 0;
-         }
-
-         template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class Iterator>
-         number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
-            import_bits_generic(
-               number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, Iterator i, Iterator j, unsigned chunk_size = 0, bool msv_first = true)
-         {
-            typename number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>::backend_type newval;
-
-            typedef typename std::iterator_traits<Iterator>::value_type       value_type;
-            typedef typename boost::make_unsigned<value_type>::type           unsigned_value_type;
-            typedef typename std::iterator_traits<Iterator>::difference_type  difference_type;
-            typedef typename boost::make_unsigned<difference_type>::type      size_type;
-            typedef typename cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>::trivial_tag tag_type;
-
-            if(!chunk_size)
-               chunk_size = std::numeric_limits<value_type>::digits;
-
-            size_type limbs = std::distance(i, j);
-            size_type bits = limbs * chunk_size;
-
-            detail::resize_to_bit_size(newval, static_cast<unsigned>(bits), tag_type());
-
-            difference_type bit_location = msv_first ? bits - chunk_size : 0;
-            difference_type bit_location_change = msv_first ? -static_cast<difference_type>(chunk_size) : chunk_size;
-
-            while(i != j)
-            {
-               detail::assign_bits(newval, static_cast<unsigned_value_type>(*i), static_cast<unsigned>(bit_location), chunk_size, tag_type());
-               ++i;
-               bit_location += bit_location_change;
-            }
-
-            newval.normalize();
-
-            val.backend().swap(newval);
-            return val;
-         }
-
-         template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
-         inline typename boost::disable_if_c<boost::multiprecision::backends::is_trivial_cpp_int<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator> >::value, number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&>::type
-            import_bits_fast(
-               number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0)
-         {
-            std::size_t byte_len = (j - i) * (chunk_size ? chunk_size / CHAR_BIT : sizeof(*i));
-            std::size_t limb_len = byte_len / sizeof(limb_type);
-            if(byte_len % sizeof(limb_type))
-               ++limb_len;
-            cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& result = val.backend();
-            result.resize(static_cast<unsigned>(limb_len), static_cast<unsigned>(limb_len));  // checked types may throw here if they're not large enough to hold the data!
-            result.limbs()[result.size() - 1] = 0u;
-            std::memcpy(result.limbs(), i, (std::min)(byte_len, result.size() * sizeof(limb_type)));
-            result.normalize(); // In case data has leading zeros.
-            return val;
-         }
-         template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
-         inline typename boost::enable_if_c<boost::multiprecision::backends::is_trivial_cpp_int<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator> >::value, number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&>::type
-            import_bits_fast(
-               number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0)
-         {
-            cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& result = val.backend();
-            std::size_t byte_len = (j - i) * (chunk_size ? chunk_size / CHAR_BIT : sizeof(*i));
-            std::size_t limb_len = byte_len / sizeof(result.limbs()[0]);
-            if(byte_len % sizeof(result.limbs()[0]))
-               ++limb_len;
-            result.limbs()[0] = 0u;
-            result.resize(static_cast<unsigned>(limb_len), static_cast<unsigned>(limb_len));  // checked types may throw here if they're not large enough to hold the data!
-            std::memcpy(result.limbs(), i, (std::min)(byte_len, result.size() * sizeof(result.limbs()[0])));
-            result.normalize(); // In case data has leading zeros.
-            return val;
-         }
-      }
+namespace multiprecision {
+
+namespace detail {
 
+template <class Backend, class Unsigned>
+void assign_bits(Backend& val, Unsigned bits, unsigned bit_location, unsigned chunk_bits, const mpl::false_& tag)
+{
+   unsigned limb  = bit_location / (sizeof(limb_type) * CHAR_BIT);
+   unsigned shift = bit_location % (sizeof(limb_type) * CHAR_BIT);
 
-      template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class Iterator>
-      inline number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
-         import_bits(
-            number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, Iterator i, Iterator j, unsigned chunk_size = 0, bool msv_first = true)
+   limb_type mask = chunk_bits >= sizeof(limb_type) * CHAR_BIT ? ~static_cast<limb_type>(0u) : (static_cast<limb_type>(1u) << chunk_bits) - 1;
+
+   limb_type value = static_cast<limb_type>(bits & mask) << shift;
+   if (value)
+   {
+      if (val.size() == limb)
       {
-         return detail::import_bits_generic(val, i, j, chunk_size, msv_first);
+         val.resize(limb + 1, limb + 1);
+         if (val.size() > limb)
+            val.limbs()[limb] = value;
       }
+      else if (val.size() > limb)
+         val.limbs()[limb] |= value;
+   }
+   if (chunk_bits > sizeof(limb_type) * CHAR_BIT - shift)
+   {
+      shift = sizeof(limb_type) * CHAR_BIT - shift;
+      chunk_bits -= shift;
+      bit_location += shift;
+      bits >>= shift;
+      if (bits)
+         assign_bits(val, bits, bit_location, chunk_bits, tag);
+   }
+}
+template <class Backend, class Unsigned>
+void assign_bits(Backend& val, Unsigned bits, unsigned bit_location, unsigned chunk_bits, const mpl::true_&)
+{
+   typedef typename Backend::local_limb_type local_limb_type;
+   //
+   // Check for possible overflow, this may trigger an exception, or have no effect
+   // depending on whether this is a checked integer or not:
+   //
+   if ((bit_location >= sizeof(local_limb_type) * CHAR_BIT) && bits)
+      val.resize(2, 2);
+   else
+   {
+      local_limb_type mask  = chunk_bits >= sizeof(local_limb_type) * CHAR_BIT ? ~static_cast<local_limb_type>(0u) : (static_cast<local_limb_type>(1u) << chunk_bits) - 1;
+      local_limb_type value = (static_cast<local_limb_type>(bits) & mask) << bit_location;
+      *val.limbs() |= value;
+      //
+      // Check for overflow bits:
+      //
+      bit_location = sizeof(local_limb_type) * CHAR_BIT - bit_location;
+      if ((bit_location < sizeof(bits) * CHAR_BIT) && (bits >>= bit_location))
+         val.resize(2, 2); // May throw!
+   }
+}
 
-      template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
-      inline number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
-         import_bits(
-            number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0, bool msv_first = true)
-      {
-#ifdef BOOST_LITTLE_ENDIAN
-         if(((chunk_size % CHAR_BIT) == 0) && !msv_first)
-            return detail::import_bits_fast(val, i, j, chunk_size);
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator>
+inline void resize_to_bit_size(cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& newval, unsigned bits, const mpl::false_&)
+{
+   unsigned limb_count = static_cast<unsigned>(bits / (sizeof(limb_type) * CHAR_BIT));
+   if (bits % (sizeof(limb_type) * CHAR_BIT))
+      ++limb_count;
+   static const unsigned max_limbs = MaxBits ? MaxBits / (CHAR_BIT * sizeof(limb_type)) + ((MaxBits % (CHAR_BIT * sizeof(limb_type))) ? 1 : 0) : (std::numeric_limits<unsigned>::max)();
+   if (limb_count > max_limbs)
+      limb_count = max_limbs;
+   newval.resize(limb_count, limb_count);
+   std::memset(newval.limbs(), 0, newval.size() * sizeof(limb_type));
+}
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator>
+inline void resize_to_bit_size(cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& newval, unsigned, const mpl::true_&)
+{
+   *newval.limbs() = 0;
+}
+
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class Iterator>
+number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
+import_bits_generic(
+    number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, Iterator i, Iterator j, unsigned chunk_size = 0, bool msv_first = true)
+{
+   typename number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>::backend_type newval;
+
+   typedef typename std::iterator_traits<Iterator>::value_type                                   value_type;
+   typedef typename boost::make_unsigned<value_type>::type                                       unsigned_value_type;
+   typedef typename std::iterator_traits<Iterator>::difference_type                              difference_type;
+   typedef typename boost::make_unsigned<difference_type>::type                                  size_type;
+   typedef typename cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>::trivial_tag tag_type;
+
+   if (!chunk_size)
+      chunk_size = std::numeric_limits<value_type>::digits;
+
+   size_type limbs = std::distance(i, j);
+   size_type bits  = limbs * chunk_size;
+
+   detail::resize_to_bit_size(newval, static_cast<unsigned>(bits), tag_type());
+
+   difference_type bit_location        = msv_first ? bits - chunk_size : 0;
+   difference_type bit_location_change = msv_first ? -static_cast<difference_type>(chunk_size) : chunk_size;
+
+   while (i != j)
+   {
+      detail::assign_bits(newval, static_cast<unsigned_value_type>(*i), static_cast<unsigned>(bit_location), chunk_size, tag_type());
+      ++i;
+      bit_location += bit_location_change;
+   }
+
+   newval.normalize();
+
+   val.backend().swap(newval);
+   return val;
+}
+
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
+inline typename boost::disable_if_c<boost::multiprecision::backends::is_trivial_cpp_int<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator> >::value, number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&>::type
+import_bits_fast(
+    number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0)
+{
+   std::size_t byte_len = (j - i) * (chunk_size ? chunk_size / CHAR_BIT : sizeof(*i));
+   std::size_t limb_len = byte_len / sizeof(limb_type);
+   if (byte_len % sizeof(limb_type))
+      ++limb_len;
+   cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& result = val.backend();
+   result.resize(static_cast<unsigned>(limb_len), static_cast<unsigned>(limb_len)); // checked types may throw here if they're not large enough to hold the data!
+   result.limbs()[result.size() - 1] = 0u;
+   std::memcpy(result.limbs(), i, (std::min)(byte_len, result.size() * sizeof(limb_type)));
+   result.normalize(); // In case data has leading zeros.
+   return val;
+}
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
+inline typename boost::enable_if_c<boost::multiprecision::backends::is_trivial_cpp_int<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator> >::value, number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&>::type
+import_bits_fast(
+    number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0)
+{
+   cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>& result   = val.backend();
+   std::size_t                                                      byte_len = (j - i) * (chunk_size ? chunk_size / CHAR_BIT : sizeof(*i));
+   std::size_t                                                      limb_len = byte_len / sizeof(result.limbs()[0]);
+   if (byte_len % sizeof(result.limbs()[0]))
+      ++limb_len;
+   result.limbs()[0] = 0u;
+   result.resize(static_cast<unsigned>(limb_len), static_cast<unsigned>(limb_len)); // checked types may throw here if they're not large enough to hold the data!
+   std::memcpy(result.limbs(), i, (std::min)(byte_len, result.size() * sizeof(result.limbs()[0])));
+   result.normalize(); // In case data has leading zeros.
+   return val;
+}
+} // namespace detail
+
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class Iterator>
+inline number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
+import_bits(
+    number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, Iterator i, Iterator j, unsigned chunk_size = 0, bool msv_first = true)
+{
+   return detail::import_bits_generic(val, i, j, chunk_size, msv_first);
+}
+
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class T>
+inline number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>&
+import_bits(
+    number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, T* i, T* j, unsigned chunk_size = 0, bool msv_first = true)
+{
+#if BOOST_ENDIAN_LITTLE_BYTE
+   if (((chunk_size % CHAR_BIT) == 0) && !msv_first)
+      return detail::import_bits_fast(val, i, j, chunk_size);
 #endif
-         return detail::import_bits_generic(val, i, j, chunk_size, msv_first);
-      }
+   return detail::import_bits_generic(val, i, j, chunk_size, msv_first);
+}
 
-      namespace detail {
-
-         template <class Backend>
-         boost::uintmax_t extract_bits(const Backend& val, unsigned location, unsigned count, const mpl::false_& tag)
-         {
-            unsigned limb = location / (sizeof(limb_type) * CHAR_BIT);
-            unsigned shift = location % (sizeof(limb_type) * CHAR_BIT);
-            boost::uintmax_t result = 0;
-            boost::uintmax_t mask = count == std::numeric_limits<boost::uintmax_t>::digits ? ~static_cast<boost::uintmax_t>(0) : (static_cast<boost::uintmax_t>(1u) << count) - 1;
-            if(count > (sizeof(limb_type) * CHAR_BIT - shift))
-            {
-               result = extract_bits(val, location + sizeof(limb_type) * CHAR_BIT - shift, count - sizeof(limb_type) * CHAR_BIT + shift, tag);
-               result <<= sizeof(limb_type) * CHAR_BIT - shift;
-            }
-            if(limb < val.size())
-               result |= (val.limbs()[limb] >> shift) & mask;
-            return result;
-         }
-
-         template <class Backend>
-         inline boost::uintmax_t extract_bits(const Backend& val, unsigned location, unsigned count, const mpl::true_&)
-         {
-            typename Backend::local_limb_type result = *val.limbs();
-            typename Backend::local_limb_type mask = count >= std::numeric_limits<typename Backend::local_limb_type>::digits ? ~static_cast<typename Backend::local_limb_type>(0) : (static_cast<typename Backend::local_limb_type>(1u) << count) - 1;
-            return (result >> location) & mask;
-         }
+namespace detail {
+
+template <class Backend>
+boost::uintmax_t extract_bits(const Backend& val, unsigned location, unsigned count, const mpl::false_& tag)
+{
+   unsigned         limb   = location / (sizeof(limb_type) * CHAR_BIT);
+   unsigned         shift  = location % (sizeof(limb_type) * CHAR_BIT);
+   boost::uintmax_t result = 0;
+   boost::uintmax_t mask   = count == std::numeric_limits<boost::uintmax_t>::digits ? ~static_cast<boost::uintmax_t>(0) : (static_cast<boost::uintmax_t>(1u) << count) - 1;
+   if (count > (sizeof(limb_type) * CHAR_BIT - shift))
+   {
+      result = extract_bits(val, location + sizeof(limb_type) * CHAR_BIT - shift, count - sizeof(limb_type) * CHAR_BIT + shift, tag);
+      result <<= sizeof(limb_type) * CHAR_BIT - shift;
+   }
+   if (limb < val.size())
+      result |= (val.limbs()[limb] >> shift) & mask;
+   return result;
+}
 
-      }
+template <class Backend>
+inline boost::uintmax_t extract_bits(const Backend& val, unsigned location, unsigned count, const mpl::true_&)
+{
+   typename Backend::local_limb_type result = *val.limbs();
+   typename Backend::local_limb_type mask   = count >= std::numeric_limits<typename Backend::local_limb_type>::digits ? ~static_cast<typename Backend::local_limb_type>(0) : (static_cast<typename Backend::local_limb_type>(1u) << count) - 1;
+   return (result >> location) & mask;
+}
 
-      template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class OutputIterator>
-      OutputIterator export_bits(
-         const number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, OutputIterator out, unsigned chunk_size, bool msv_first = true)
-      {
+} // namespace detail
+
+template <unsigned MinBits, unsigned MaxBits, cpp_integer_type SignType, cpp_int_check_type Checked, class Allocator, expression_template_option ExpressionTemplates, class OutputIterator>
+OutputIterator export_bits(
+    const number<cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>, ExpressionTemplates>& val, OutputIterator out, unsigned chunk_size, bool msv_first = true)
+{
 #ifdef _MSC_VER
 #pragma warning(push)
-#pragma warning(disable:4244)
+#pragma warning(disable : 4244)
 #endif
-         typedef typename cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>::trivial_tag tag_type;
-         if(!val)
-         {
-            *out = 0;
-            ++out;
-            return out;
-         }
-         unsigned bitcount = boost::multiprecision::backends::eval_msb_imp(val.backend()) + 1;
-         unsigned chunks = bitcount / chunk_size;
-         if(bitcount % chunk_size)
-            ++chunks;
-
-         int bit_location = msv_first ? bitcount - chunk_size : 0;
-         int bit_step = msv_first ? -static_cast<int>(chunk_size) : chunk_size;
-         while(bit_location % bit_step) ++bit_location;
-
-         do
-         {
-            *out = detail::extract_bits(val.backend(), bit_location, chunk_size, tag_type());
-            ++out;
-            bit_location += bit_step;
-         } while((bit_location >= 0) && (bit_location < (int)bitcount));
-
-         return out;
+   typedef typename cpp_int_backend<MinBits, MaxBits, SignType, Checked, Allocator>::trivial_tag tag_type;
+   if (!val)
+   {
+      *out = 0;
+      ++out;
+      return out;
+   }
+   unsigned bitcount = boost::multiprecision::backends::eval_msb_imp(val.backend()) + 1;
+   unsigned chunks   = bitcount / chunk_size;
+   if (bitcount % chunk_size)
+      ++chunks;
+
+   int bit_location = msv_first ? bitcount - chunk_size : 0;
+   int bit_step     = msv_first ? -static_cast<int>(chunk_size) : chunk_size;
+   while (bit_location % bit_step)
+      ++bit_location;
+
+   do
+   {
+      *out = detail::extract_bits(val.backend(), bit_location, chunk_size, tag_type());
+      ++out;
+      bit_location += bit_step;
+   } while ((bit_location >= 0) && (bit_location < (int)bitcount));
+
+   return out;
 #ifdef _MSC_VER
 #pragma warning(pop)
 #endif
-      }
-
-   }
 }
 
-
+}
+} // namespace boost::multiprecision
 
 #endif // BOOST_MP_CPP_INT_IMPORT_EXPORT_HPP
-